Complexity and Simplicity of Org Dynamics Christian Lemp

and i'm going to talk about complexity and complex systems and ways that we can think about modeling uh from nature to inspire how we think about dowse and organizations so really good timing because we just had two really cool talks talking about incentives coordination of lots of different distributed resources and so today i'm going to go into some ways that we can start thinking about modeling in that way a little about me my name is christian i co-founded a dao called diamond dao we're focused on building out a curated knowledge graph to help people find interesting frontier communities across the space but we start when you think about systems i always go and i go back to nature which is the ultimate sort of distributed decentralized uh system that you know we can only really begin to imagine how does everything come together right so we think about cells starting at the kind of most individual basic level cells are sitting there processing information coordinating with each other and yet kind of emerge up as a complex system into ourselves and here we are going then beyond ourselves we think about just beyond the organism we can think about organisms coming together and building flocks right so here's some amazing uh photographs from a photographer who takes time series of birds flocking together and we can see that there's like some sense of order that's going on here but there's no one individual who's in control right there's no boss saying hey birds go this way they're all kind of flying around and i think like this is a really nice analogy to think about taos right or at least in our sort of uh purest form of decentralization who's the boss how do they come together well there's definitely structure and order there but like how do we find it and the answer there right so how do we go from sort of organization to simulation and understand kind of as a computer how can we test our intuitions through models right the answer there the approach that's used to understand nature are agent-based models right so an agent is one individual and we can find with very simple rules organized in the right way we can create amazing and sort of uh emergent behavior so what you're seeing in the background there is a simulation of birds flocking together and there's only three parameters to this model right so it's not millions of data points that's needed we don't need to go out there and study like how birds migrate across you know every different ecosystem with just three sort of parameters tuned the right way and creatively applied to the system we can start to generate behavior that looks real and get an intuition about what's going on with each individual individual agent here so from there i take it and i think what can we apply then about our human organizations right like what are the simple rules and behaviors that sort of govern how we interact right so we saw in like earlier presentation there's like thousands of people in a discord there's like lots of governance going on there's people making choices you know we're part of these systems but we really have like very limited knowledge about what each other is thinking each other's motivations biases perspective all of these things that we bring to the organization can we start to think about maybe applying simple rules to get some intuition about what's actually going on right so the framework to think about if you're interested in applying this kind of mental model to your organization is think what's the micro what's the macro what's the interaction both within that level and then across levels right so agents right these might be members of our dao well how do we actually interact like what information specifically are we sharing with each other right and when we interpret that information from each other what does that do to our own perspective right so we think we can simulate over time like we saw with the birds in the prior slide instead we're going to be simulating what happens to group behaviors over time right so what choices are being made at each step right how can we kind of formalize that with a uh the model what knowledge do we have about each other when we're making these right these are all computer choices right so it's about being sort of creative and applying perspective to what's going on and then formalizing it into some sort of model to think like what happens if i play this incentive out for 10 years what's going to actually happen to my organization if things play out the way that i organize it right and then each agent interacts with the environment right so our environment we kind of uh decide what that is right in the dow that's kind of the organization right like that's the treasury for example that's the network structure that's underlying right so there's simple rules that individual agents are making and then that's affecting the environment which then creates a feedback loop back to the individuals and then it carries on and we have this like complex dynamic system going on right so if you want to get started and you want to use this kind of modeling you don't need to be quantitative to do this and this is what i like about agent-based models is you just have to have perspective intuition about what's going on right so you start by naming the parts right so diamond dial like i mentioned we're on dow house we use the molok dao framework and what are the parts right of this model that i'm going to show you the results of in a second one there's agents those are dow members so choice i made in this simple model for this presentation everybody's got the same number of shares right there's a proposal i was thinking about proposals and like what does a proposal actually mean means that you're like accepting something that's in delivering value back into the dow in some way there's like kind of a couple dimensions right so proposal happens it could be like an engineering innovation is one dimension and right there's some value it's going to add in that sense there's maybe like really cool community good vibes that's going to happen right so maybe that's another dimension right so in this model i chose like three dimensions of value that a proposal will have then the environment small world network so like people at this conference you probably have met people that you didn't know you had a common connection you're like wow that's such a small world there's actually a formal way there's a model there's a network model we can use to generate that same sort of uh you know small world connection six degrees of separation type of phenomenon and then there's like time and randomness it's like what's random in the model so for the one that i built here what's random is in this consensus building period when voting is open like who do you talk to to like share perspective on how valuable you think this proposal actually is right so that's random there's a randomness going on there and then time each each time step is one day and so kind of some like simple results but to kind of generate how to think about like what's going on is i said well what happens if i simulate this like a whole bunch of times but then i increase the like the number of people in the dow but i keep their kind of close network relatively the same so like each down member maybe has like three close friends that they do this uh you know checking of like consensus building with what's the value of this proposal and each agent's got like this threshold if they think it's valuable beyond this threshold they're going to vote yes if not they're going to vote no what i want to understand is like if i increase the size of this dow what does that do to like whether proposals pass or not and so it was kind of interesting i had no idea what i was going to see when i ran this but i found that like up to 20 members all of a sudden we kind of crater into this like almost nothing's getting passed so like well that's kind of interesting right you start with this like small group everybody's relatively aligned starts to grow you go through this like trough of like consensus building it's really hard you get through that and then all of a sudden like consensus is easier reached again it kind of like mirrors experience in my own like life in the space of like seeing what happens as the community scales there's usually like this kind of like hard period where it's like medium size and people are forming so i thought this result was kind of interesting then for the second example i said okay what happens if we start with like a really big community and then instead of like changing how big that community is change like how many close connections uh each member of the dao is going to have right so start with like just two close friends who you build consensus with and you said hey this proposal's worth passing it's really valuable for xyz reasons well what happens if you start taking more and more perspectives from more and more people as you're trying to get a vote pass or a proposal passed i saw that like this was i think for like 30 30 member dao once you get again it like eight connections all of a sudden like consensus craters for whatever reason i don't know why right so this is the thing it's about identifying structure so identifying pattern i don't exactly know why this happens and then beyond that if you start to like broaden out your group who you build consensus with all of a sudden it's easier reached kind of interesting right kind of maybe you can think of like experiences in your own dao that might kind of relate or maybe not and you've got a different perspective on kind of what model should be used so if you're really into the code and you really want to like explore this then you can go this is on github so check it out take a picture start it fork it play around right this is very simple but i'd love to see new perspectives in how to build this out and make it maybe more realistic cool ideas i'll just seed some ideas into the crowd and see where this goes you could refactor the code so block science is a really cool package called cadcad for agent based modeling and complex adaptive systems i didn't use that because i'm not that familiar with it yet malek v3 is coming out has all kinds of new rules and all sorts of new incentives maybe you could make an agent-based model to see what those new rules can do to your community you could like simulate and visualize it and make it like more compelling and then if you're qualitative you're like i'm not really a coder but i really like being in the dallas space and this is fascinating again you could for example maybe you can create a survey when i was talking about like the value of these different dimensions of a proposal maybe you could come up with a system to like map those to real numbers and then we could see like does their simulation match reality or you could create a whole new story altogether right i just came up with a story of people coming together in a small world building consensus they have a threshold and then if it passes a threshold then they vote yes that's a story i made that up you can make up your own story figure out a way or you know connect with somebody who likes to code put that together and see you know where does my intuition take me so thank you all for this extremely quick talk into agent-based models complex systems follow me on twitter i'd love to see people uh with ideas you know i love complexity doing a phd in this it's like you know fascinating for me to talk about this so always welcome and i think we've got yeah we have five minutes for a couple of questions